{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:NNJ7UUWWIN5ZSL5ZB5OM6NTPZO","short_pith_number":"pith:NNJ7UUWW","schema_version":"1.0","canonical_sha256":"6b53fa52d6437b992fb90f5ccf366fcb9eef29ea936a13d27c3f80c5756c3abb","source":{"kind":"arxiv","id":"2203.01517","version":2},"attestation_state":"computed","paper":{"title":"Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chelsea Finn, Christopher R\\'e, Hongyang R. Zhang, Michael Zhang, Nimit S. Sohoni","submitted_at":"2022-03-03T05:03:28Z","abstract_excerpt":"Spurious correlations pose a major challenge for robust machine learning. Models trained with empirical risk minimization (ERM) may learn to rely on correlations between class labels and spurious attributes, leading to poor performance on data groups without these correlations. This is particularly challenging to address when spurious attribute labels are unavailable. To improve worst-group performance on spuriously correlated data without training attribute labels, we propose Correct-N-Contrast (CNC), a contrastive approach to directly learn representations robust to spurious correlations. As"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2203.01517","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-03-03T05:03:28Z","cross_cats_sorted":[],"title_canon_sha256":"731575e796fc5993a4117908151609426926afbfb1e8b0585e60f3e6de405f9f","abstract_canon_sha256":"199561af05e52c270fbd8578fa3013880d904b9bd9ae82b333afd1e6280b1278"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:47:24.723760Z","signature_b64":"bbZn5s0SYJZlwyuB6m7pWieh770na36V+6nTrkcXVlKnYa7tC2ClSDL91v5NNtKqiXNSeRYz3s8rqazSX8ewDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6b53fa52d6437b992fb90f5ccf366fcb9eef29ea936a13d27c3f80c5756c3abb","last_reissued_at":"2026-07-05T09:47:24.723284Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:47:24.723284Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Correct-N-Contrast: A Contrastive Approach for Improving Robustness to Spurious Correlations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chelsea Finn, Christopher R\\'e, Hongyang R. Zhang, Michael Zhang, Nimit S. Sohoni","submitted_at":"2022-03-03T05:03:28Z","abstract_excerpt":"Spurious correlations pose a major challenge for robust machine learning. Models trained with empirical risk minimization (ERM) may learn to rely on correlations between class labels and spurious attributes, leading to poor performance on data groups without these correlations. This is particularly challenging to address when spurious attribute labels are unavailable. To improve worst-group performance on spuriously correlated data without training attribute labels, we propose Correct-N-Contrast (CNC), a contrastive approach to directly learn representations robust to spurious correlations. As"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.01517","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2203.01517/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2203.01517","created_at":"2026-07-05T09:47:24.723346+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.01517v2","created_at":"2026-07-05T09:47:24.723346+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.01517","created_at":"2026-07-05T09:47:24.723346+00:00"},{"alias_kind":"pith_short_12","alias_value":"NNJ7UUWWIN5Z","created_at":"2026-07-05T09:47:24.723346+00:00"},{"alias_kind":"pith_short_16","alias_value":"NNJ7UUWWIN5ZSL5Z","created_at":"2026-07-05T09:47:24.723346+00:00"},{"alias_kind":"pith_short_8","alias_value":"NNJ7UUWW","created_at":"2026-07-05T09:47:24.723346+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24161","citing_title":"Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglement","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07451","citing_title":"TEVI: Text-Conditioned Editing of Visual Representations via Sparse Autoencoders for Improved Vision-Language Alignment","ref_index":189,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02830","citing_title":"Mitigating Spurious Correlations with Memorization-Guided Dataset De-Biasing","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03148","citing_title":"$A^2$: Smaller Self-Supervised ViTs Localize Better than Larger Ones","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2603.08639","citing_title":"UNBOX: Unveiling Black-box visual models with Natural-language","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO","json":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO.json","graph_json":"https://pith.science/api/pith-number/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/graph.json","events_json":"https://pith.science/api/pith-number/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/events.json","paper":"https://pith.science/paper/NNJ7UUWW"},"agent_actions":{"view_html":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO","download_json":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO.json","view_paper":"https://pith.science/paper/NNJ7UUWW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.01517&json=true","fetch_graph":"https://pith.science/api/pith-number/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/graph.json","fetch_events":"https://pith.science/api/pith-number/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/action/storage_attestation","attest_author":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/action/author_attestation","sign_citation":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/action/citation_signature","submit_replication":"https://pith.science/pith/NNJ7UUWWIN5ZSL5ZB5OM6NTPZO/action/replication_record"}},"created_at":"2026-07-05T09:47:24.723346+00:00","updated_at":"2026-07-05T09:47:24.723346+00:00"}